sources (version 1.5)

• the top level processing calling tree
• the code itself

The positions and extraction radii for the sources are determined in a four step process.

1. Smooth the summed sky and exposure maps with the vsmooth STOOL. This does a variable size boxcar smooth which uses the minimum square box of pixels which contains at least 50 events. The same size box is applied to the counts image and to the exposure map and the results are divided to produce a smoothed, exposure-corrected image. Pixels with 0 exposure are assigned a pixel value of 0. Note that each combination of instrument (SIS or GIS) and image resolution (256 or possibly 64 for the GIS), has three images corresponding to lo (<2keV), hi (>2keV) and all energy bands.
2. Detect sources in the smoothed images. This is done using the ascasource STOOL which reports the center pixel of each source s well as a suggested extraction radius and a maximum extraction radius to avoid confusion with other sources. Sources are detected with the following algorithm:
   1. Locate the 20 brightest local peaks in the image. A local peak is a pixel which is as bright or brighter than the eight pixels surrounding it.
   2. Examine the peaks to see if they are distinct and significant. Choose the brightest peak. For each of the other peaks, compare the minimum intensity along the line segment connecting the peaks to the intensity of the lower peak. If the lower peak is less than 2.5 times brighter than the trough separating it from the brightest peak, delete the lower peak. repeat this for the second brightest surviving peak, etc.
   3. Determine the maximum extraction radii. For each source calculate the distance to the the lowest point along a straight line to each other source. Set the maximum radius to the nearest such trough.
   4. Determine the radial distribution of azimuthally averaged intensity around each source. Pixels with zero intensity are not included in the average. The lowest intensity in the radial distrubution is taken as the background flux.
   5. Delete all sources whose peak intensity is less than four times the background intensity.
   6. Determine the "suggested" extraction radius. This is set to the minimum radius at which the azimuthally averaged intensity is more than 1/5 of the peak intensity and more than twice the background intensity. The suggested radius is never greater than the maximum radius.
3. Determine the extraction radius for each source. The default extraction radius is 6 arcmin for GIS and 4 arcmin for SIS. If the suggested radius is less than the default radius (true for point sources) than the default radius is used. Otherwise use the suggested radius. In no case is the extraction radius larger than the maximum radius.
4. Reconcile the sources detected in the three energy bands. and eliminate duplicates. This is done by giving a list of all source positions and radii to the sift_sources STOOL. The sources are listed in order of decreasing intensity with all of the "all" energy bad sources listed first, follwed by the "hi" energy sources, and the "lo" energy sources. Sift_sources does the following:
   1. Take the first source in the list.
   2. Examine all the other sources in the list.
   3. If the second source is within the extraction radius of the first source, eliminate it.
   4. If the extraction radii of the two sources overlap, reduce the radius of the second source so that they just touch.
   5. Repeat these steps for each of the other remaining sources

FTOOLS called:

1. fcreate
2. fdump
3. fmodhead
4. setup_parfile

1. ¸
2. ascasource
3. colorpic
4. counts_in_circle
5. equals
6. g
7. hïM8
8. sift_sources
9. vsmooth

1. exception
2. ftool_test
3. generate_filename
4. log_entry
5. milestone
6. parse_filename
7. read_bintable
8. read_fits_keyword
9. read_parfile
10. setup_parfile
11. sky2det
12. stool_test